Re: design question -- localized metadata?

From: Karl Ove Hufthammer (
Date: Wed May 15 2002 - 10:22:29 EDT

  • Next message: Rui Miguel Silva Seabra: "Re: metadata namespace screw cases and DC/RDF"

    Dom Lachowicz <> wrote in

    > These guys need to get a good grip on semantics and the proper
    > usage of several English words.

    They're mostly librarians -- experts at classifications. You can
    trust they have a better grip on semantics than any of us.

    > They're using words that are
    > related, but not synonyms as though they were synonyms.

    No, they're not. I'll explain below.

    > Subject != Keywords. Even "Subjects Covered or Topics" !=
    > Keywords. Maybe they are similar but it's still not (nearly)
    > correct. "Subject" is much closer to "Description"
    > semantically, but Descriptions are much longer than Subjects
    > in practice (where description = a 1+ sentence summary). So
    > what they're saying is:

    No, they're not. The name of the element is 'Subject and
    Keywords'. Most classifcation schemes, such as 'Dewey Decimal
    Classification' and the 'Library of Congress Classification'
    classify text by 'subjects', often in an hierarchy. On the other
    hand, Web pages are classified (by the authors) using 'keywords'
    or 'key phrases' (seldomly from an controlled vocabulary,

    Now the Dublin Core people *combined* subjects and keywords / key
    phrases into *one* element, which they named 'Subject and
    Keywords'. But the *identifier* for this element (which has to
    consist of only one word ) was named 'Subject', which they defined
    as "The topic of the content of the resource."

    With the comment:

                   Typically, a Subject will be expressed as keywords,
                   key phrases or classification codes that describe a
                   topic of the resource.

    Everything clear now?

    Karl Ove Hufthammer

    This archive was generated by hypermail 2.1.4 : Wed May 15 2002 - 10:26:44 EDT